Fatalities in Swedish fire-related car crashes from a toxicologic perspective

Abstract

Objective: Vehicle materials developments raise concerns about new patterns of vehicle fire toxic gas emissions. This study aimed to describe toxicologic components in a recent material of fatal car crashes on Swedish roads in which the vehicle caught fire and compare the results to a previous material.

Methods: Retrospective registry study. All fatal car crashes with fire in Sweden 2009–2018 were extracted from the Swedish Transport Administration’s In-Depth Studies Database and compared with an earlier study of the time period 1998–2008.

Results: A total of 79 crashes and 94 fatalities were included. Carbon monoxide (COHb) blood levels >10% were found in 13 cases. Hydrogen cyanide (HCN) blood levels 0.1–1.7 µg/g were found in 10 cases. In 29 of the cases the person had a blood alcohol level (BAC) >0.2‰, which is the legal driving limit in Sweden. A total of 15 people died due to burn injuries and 2 individuals died due to toxic gas emissions without any other fatal traumatic injury. Total number of deaths in fire-related crashes halved from 181 (1998–2008) to 94 (2009–2018) but the percentage of fatalities in burning vehicles was unaltered (5% vs. 6%). The proportion of fatalities with HCN in the blood increased from 2% between 1998–2008 to 10% during 2009–2018 (p = 0.006). The age of the car involved in a crash increased by 0.26 years per calendar year (p = 0.001).

Conclusions: The proportion of fatalities with measured levels of HCN in the blood has increased. Eleven of the 15 burn injury fatalities had high levels of alcohol, HCN, or COHb, possibly contributing to an inability to leave a burning vehicle. Faster rescue brought by improved specific education and training of ambulance and rescue services personnel may be of future importance, as may on-scene antidote administration and revised regulations of vehicle flammability.

Keywords:

• Toxic gas emissions
• vehicle fire
• car crashes
• hydrogen cyanide
• carbon monoxide

Introduction

During the last decade, technological progress in motor vehicle design and manufacturing has moved forward quickly. The ambition to lower vehicle emissions has led to the implementation of alternative fuel sources (Larsson et al. 2013; Lecocq et al. 2012), and construction materials such as carbon fiber, polyurethane and magnesium alloys, as well as chemical components such as new air-cooling systems (Mckenna and Hull 2016). When burning, the decomposition of polyurethane creates carbon monoxide (CO) and hydrogen cyanide (HCN) (Mckenna and Hull 2016). Acute CO poisoning can vary in clinical representation, but the observed symptoms often correlate with the percentage of carboxyhemoglobin (COHb) levels observed in the blood. When levels rise over 10%, symptoms such as dyspnea, headache and angina can be observed. When they rise over 20%, symptoms increase with nausea, vomiting, difficulty with concentration as well as fatigability. The higher the percentage of COHb, the more severe the symptoms become (Llano and Raffin 1990). HCN poisoning leads to cellular hypoxia and lactic acidosis by a noncompetitive inhibition of the mitochondrial Cytochrome C oxidase activity (Baud et al. 1991). HCN poisoning, as well as CO poisoning, can be reversed. While CO is treated mainly with 100% oxygen and symptomatic treatment of the airways, HCN poisoning can be treated with an antidote such as hydroxocobalamin (Cyanokit®). The antidote should be administered intravenously as soon as possible (Hall et al. 2009), preferably on the crash site. Truchot et al. (2018) used a fire tunnel to analyze the emissions from a burning vehicle, finding that not only HCN but also hydrogen chloride (HCl) and hydrogen fluoride (HF) were produced at significant levels, with HCN being produced continuously while HCl and HF had peaks of emissions during the fire. Lönnermark and Blomqvist (2006) showed that some of the highest HCN-producing materials were door panels, carpets, dashboards, and polyurethane upholstered seats.

A previous study by our group (Viklund et al. 2013) analyzed causal factors behind deadly outcomes in car crashes between the years 1998–2008 where the vehicle caught fire, using the Swedish Transport Administration In-Depth Studies Database. This registry, commissioned by the government as a tool to improve road safety, has collected data from all known fatal road traffic incidents in Sweden since 1998. Data sources include the police, rescue services, ambulances, autopsy reports, witnesses, and the Swedish Transport Administration’s own investigators. Nearly all victims of fatal car crashes in Sweden are subject to an autopsy and the primary cause of death is presented in the death certificate (Socialdepartementet 1995). Autopsy toxicological analyses for burn victims typically include blood alcohol concentration (BAC), CO concentration (COHb), and sometimes HCN concentration. Our 1998–2008 study concluded that a third of victims died due to burn injuries and/or smoke inhalation and pointed to a need for more fire-resistant vehicle designs. Today, ten years on, it would seem of interest to repeat the investigation, considering the vehicle materials developments that have taken place.

This study aimed to analyze fatalities in Swedish fire-related car crashes from a toxicological perspective between 2009–2018, and relate them to the corresponding fatality patterns for the previous ten-years period of 1998–2008. The study was approved in advance by the Swedish Ethical Review Authority (2019-06137).

Methods

All known fatal car crashes with fire in Sweden 2009–2018 were retrospectively extracted from the Swedish Transport Administration’s In-Depth Studies Database. A car was defined as having a weight <3,500 kg and being allowed to be driven with a Swedish B traffic license. Ongoing police investigations were excluded due to limitations in data availability on crash and cause of death. We also excluded fatalities in other vehicles such as trucks and motorcycles. The In-Depth study Database was found to contain 3,202 reports of fatal road traffic incidents in the period of 2009–2018. One report may contain several fatalities. The variables are: Gender, age, model of the vehicle, age of the vehicle, if seatbelts were used, the cause of fire, fuel source, primary fire source, primary cause of death, toxicological analysis, the extent of the fire, and position inside the vehicle. The sequence of events is also described. All reports were reviewed in chronological order from 2009 onward. In total, 79 incidents with 94 fatalities were found to meet our inclusion criteria, and were included for study. Fatalities were divided into groups by levels of COHb, if HCN was found in the blood, and primary cause of death. The COHb-level was set to 10% or higher since that is the level where symptoms become noticeable. These groups were chosen and compared to evaluate if the change in vehicles between two study periods had an impact on the primary cause of death. For the 1998–2008 time period, the results from our previous study (Viklund et al. 2013) were used. The data from this study also gave information regarding HCN and COHb levels which we compared our data with.

Basic descriptive statistics were used to describe population variables in 1998–2008 and 2009–2018. The test for 2-sample Poisson rates was used to test yearly death crash intensity differences between the study periods 1998–2008 and 2009–2018. Fisher’s exact test was used to evaluate if the proportion of deaths by burn injuries and the proportion of deaths with high HCN blood content have changed between the study periods. Linear regression was used to assess the change in age of the crashed car as a function of time and a t-test was used to evaluate the mean age difference of drivers between study periods. Association between fatalities due to burn injuries and high BAC levels were evaluated with Fishers exact test. All tests carried out were two sided and a significance level of 5% was used. Analyses were performed using Minitab version 19.2 Statistical Software.

Results

Characteristics of the crashes

Between 2009 and 2018, 79 different car crashes were identified in which a “B-class” vehicle caught fire and 94 people died as a result of these incidents. Single car crashes resulted in 68 (72%) of the fatalities in this group, whereas 26 (27%) were due to a collision with another vehicle. There were 21 (22%) individuals that died in crashes involving a heavy truck as the second party. Of the 94 total fatalities, 74 (78%) were drivers, 13 (13%) were front-seat passengers, and three (3%) were rear-seat passengers. In four cases (4%), the seat position was unknown. Table 1 shows the distribution of age and gender in the crashes. The death rate was highest for the age group 20–29 years at 37 (39%) casualties, as seen in Table 1. There were 18 (19%) individuals confirmed using a seat belt, and at least 21 (22%) people did not. There was a high number in which it was not stated if a seat belt had been in use or not; 54 (57%) cases in total. This was due to difficulties in the technical investigation. In one case, which was a confirmed suicide, the seat belt was found to be wound around the neck of the deceased person. Age and sex distribution of the fatalities in B-class vehicles between 2009–2018 are shown in Table 1.

Table 1. Age and sex distribution of the fatalities in B-class vehicles between 2009–2018.

Age	Female		Male		Total
(years)	N = 16		N = 78		N = 94
0–9	–	–	–	–	–	–
10–19	3	19%	11	14%	14	15%
20–29	7	44%	30	38%	37	39%
30–39	–	–	10	13%	10	11%
40–49	1	6%	8	10%	9	10%
50–59	2	12%	9	12%	11	12%
60–69	1	6%	6	8%	7	7%
70+	2	12%	4	5%	6	6%

In more than half of vehicle crashes (58%), the source of the fire was not stated. It was reported that the fire started in the engine compartment in 21 (27%) of the cases. The fire started in the fuel tank in nine (11%) of the vehicles. In one case, the fire started in the screen wash fluid container, and in another, the car exploded because of a fire in a nitrous oxide system (used in illegal street racing). Since one of the vehicles was not subject to a technical investigation, the source of the fire remained unidentified. Sixty-one vehicles had only gasoline as fuel, seven had diesel, one had gasoline/ethanol, and one car was a gasoline/Li-ion battery hybrid. In nine cases, the fuel source was unknown.

Cause of death

The autopsy report showed that 73 (78%) died from traumatic injuries and 15 (16%) due to burn injuries. There were two (2%) individuals that died due to toxic gas emissions. In four (4%) cases, an autopsy was not implemented (no data stated why not). The two casualties that were judged to have died of toxic gas emissions tested positive for HCN and had high levels of COHb (>25%). Both were males in their early 20s and also had a BAC level at the legal driving limit (Table 2). There were 17 (18%) fatalities in 16 different crashes that were classified as a suicide in the autopsy report. In seven of these crashes, the vehicle drove over to the lane of oncoming traffic and crashed into a heavy truck. In the nine other cases, the person drove straight into a fixed object such as a tree, a mountain wall, or a bridge without any evidence of braking. In 9 (60%) of 15 cases where the primary cause of death was an injury caused by burns, the individuals died at the scene of the crash, and before rescue services could remove them from the vehicle.

Table 2. Blood alcohol levels (BAC) in all fatalities.

	Total		HCN		HbCO ≥ 10%
BAC level	N = 94		N = 10		N = 13
0	55	59%	4	40%	6	46%
<0.2‰	2	2%	–	–	–	–
0.2–1.0‰	9	10%	4	40%	3	23%
>1.0–2.0‰	17	18%	1	10%	3	23%
>2.0‰	3	3%	1	10%	1	8%
Unknown	8	9%	–	–	–	–

The second and third column is the BAC levels in the cases with HCN-levels, and levels of COHb ≥ 10%. The numbers are non-cumulative.

Toxic gas emissions

Out of 94 cases, 10 tested positively for HCN in the blood. The levels ranged from 0.1 to 1.7 µg/g. Out of these 10 cases, 5 had COHb ≥ 10%. In all these cases, the car models ranged from 1996–2010. In nine vehicles the fuel was gasoline and in one case the fuel source was not registered in the database. The causes of death are stated in Table 3. The autopsies showed that 8 of the casualties were also affected by other substances, as shown in Table 4. There were two cases where the chemical autopsy report showed solely HCN and COHb in the blood. In one case, a person died due to traumatic injuries and in another case one died due to a heart attack. In 13 of the 94 cases, COHb levels were 10% or higher in the blood. The levels ranged from 10% to 50%. The car models ranged from 1993–2011. Primary cause of death in the cases with detected HCN levels or levels of COHb ≥ 10% are shown in Table 3.

Table 3. Primary cause of death in the cases with detected HCN levels or levels of COHb ≥ 10%.

	HCN		COHb
Primary cause of death	N = 10		N = 13
Traumatic injuries	3	30%	4	31%
Burn injuries	4	40%	6	46%
Toxic gas emissions	2	20%	2	15%
Cardiovascular	1	10%	1	8%

Some of the cases are included in both groups 5 out of 10 cases with HCN in the blood had COHb ≥ 10%.

Table 4. Cases with other substances in individuals with blood levels of HCN or COHb ≥ 10%.

Substance	HCN N = 10	COHb ≥ 10% N = 13
THC	1	2
SSRI	–	1
Opioids	1	–
Benzodiazepines	2	2
Amfetamine	1	1
Naproxen	1	–
Alcohol	6	8

Different substances may be found in the same individual. SSRI (selective serotonin reuptake inhibitor) is a class of antidepressants. THC (tetrahydrocannabinol) is the active substance in marijuana.

Out of the 15 persons (three female and 12 male) who died due to burn injuries, six had a COHb of 10% or higher in the blood, and four individuals had HCN levels of 0.1 µg/g blood or higher. Blood levels of alcohol were found in seven individuals. There was no significant association between fatalities due to burn injuries and driving under the influence of alcohol defined as any alcohol in the blood (p = 0.76). Cases with other substances in individuals with blood levels of HCN or COHb ≥ 10% are shown in Table 4.

Comparison with the period 1998–2008

During 1998–2008, there were a total of 133 crashes with a mean number of 12.1 crashes per year, and in 2009–2018, there were a total of 79 crashes, 7.9 per year. The difference in crash intensity between the study periods is significant (p = 0.003). Fire-related number of fatalities halved from 181 (1998–2008) to 94 (2009–2018) people in total, but the percentage of fire related deaths was about the same (5% and 6% respectively), as the total passenger car deaths have been reduced substantially. The percentage of deaths from burn injuries decreased significantly from 30% (55 out of 181) of all fatally injured, between 1998–2008, to 16% (15 out of 94) between 2009–2018 (p = 0.009). The proportion of fatalities with documented levels of HCN in the blood increased from 2% (4 out 181) between 1998–2008 (Viklund et al. 2013) to 10% (10 out of 94) during 2009–2018 (p = 0.006). The average age of the car involved in a crash was 10 years between 1998–2008 and increased to 12 years between the period of 2009–2018 (p = 0.017). The age of the car involved in a crash increased by 0.26 years per calendar year (p = 0.001). There was no significant change in the age of the driver between the study periods (p = 0.97). Between 1998–2008 the average age of the driver was 33 and between 2009–2018 the average age was 35.

Discussion

Toxic emission and burn injuries

The proportion of fatalities with documented HCN in the blood increased fivefold compared to the period of 1998–2008. In this study, concentrations of HCN in blood ranged from 0.1 to 1.7 µg/g. According to the Swedish Toxicology Information Center (Giftinformationscentralen 2020), the lethal concentration of HCN in blood postmortem is a minimum of 1.5 µg/g. However, the half-life of HCN in the blood is approximately 1 hour and it disappears quickly postmortem (Baud et al. 1991). The actual levels present at the time of death depend on several factors such as when the blood sample was taken postmortem, and how the cadaver, as well as the blood sample, were stored (Mcallister et al. 2008). The number of false-negative cases is unknown, and the actual levels of HCN at the time of death are consequently difficult to estimate. The lowest value (0.1 µg/g) of our found HCN blood concentrations is still significantly higher than what has been reported in tobacco smokers (Chaturvedi et al. 2001). There were 15 people who died due to burn injuries, without other fatal injuries. In 11 of these cases, either high levels of alcohol, HCN, or COHb ≥ 10% were found in the blood. COHb and HCN in the blood are signs of smoke inhalation. COHb ≥ 10% is a potentially toxic level (Llano and Raffin 1990). This may mean that in a majority of cases (11/15), crash victims were alive after the crash and during the subsequent fire. Inhaling CO, which has >200 times greater affinity for hemoglobin than oxygen, reduces oxygen transport capacity. Adding the effect of HCN, which blocks energy-producing mitochondria in the cells, will severely compromise oxygen utilization (Fortin et al. 2010). When inhaling CO or HCN, one of the first symptoms is dizziness and confusion (Baud 2007). Many of the individuals affected by HCN or COHb also had other substances in the blood. Benzodiazepines have a sedative effect (Buffett-Jerrott and Stewart 2002) and THC (tetrahydrocannabinol) has a negative effect on neurocognitive function (Crane et al. 2013). Alcohol is known to induce both motor and cognitive impairment (Hanchar et al. 2005). Given that the examinations were retrospective, it is impossible to establish a clear causality from accumulation of HCN and CO to the fatalities, but a contributive effect of these two combustion gases (by themselves or as an added factor on top of an elevated blood alcohol level) cannot be ruled out, including a possibly lowered capacity to exit the vehicle. In Sweden, blood levels of HCN and CO are tested during autopsy but, in comparison to CO, there is no rapid method of detection of HCN at the scene (Fortin et al. 2010). Previous studies (Truchot et al. 2018; Lönnermark and Blomqvist 2006) have also shown that both hydrogen chloride (HCl) and hydrogen fluoride (HF) are generated during vehicle fires. Since exposure to these substances may increase as electric vehicles with Li-ion batteries become more widespread, development of more comprehensive blood tests is suggested.

Vehicle materials and fuel sources

Out of the 79 crashes, diesel vehicles accounted for only 7 (9%), to be compared with 25% in the car fleet (Transport Analysis 2020). Further studies are needed to understand this difference; are diesel-driven vehicles less prone to catch fire in a crash? Since the source of the fire was unknown in more than half of all the cases, a more developed technical investigation could help to clarify if some vehicle materials or fuel sources are more prone to catch fire than others. In the United States, the Federal Motor Vehicle Safety Standard (FMVSS) has regulations as to how flammable a vehicle can be. Europe does not, to our knowledge, have an equivalent regulation. By including an assessment of a vehicle’s fire resistance in customer testing programs, for example in EuroNCAP, an incentive for developing less post-crash fire prone vehicles may be created. Electric vehicles and electric hybrids currently represent 2.5% of the total vehicle fleet (Transport Analysis 2020). Although there is, so far, no documented case of a fully electric vehicle when it comes to fatal, in-traffic crashes, the number of vehicles is increasing dramatically. The consequences of such a fire could potentially be more severe if the Li-ion battery combusts (Lecocq et al. 2012). The present results are suggested to be followed up in future studies to evaluate incidents in electric fleets. When compiling the data from the two studies, the age of the crashed vehicle increased by 0.26 years per calendar year. This would seem to suggest that even if new models of vehicles coming out of the factories should become safer regarding fire-related crashes, real world benefits may be delayed.

Suicide

In 17 of the cases, the report shows a suspected suicide. In more than half of the cases, the vehicle crashed into a heavy truck. While suicide is not the core subject matter of the present study, this result is noteworthy since it involves almost 20% of the cases. The Swedish Transport Administration has, since 1997, worked in line with the road traffic safety project "Vision Zero" aiming for a highway system with no fatalities or serious injuries. In the work toward this goal, a suicide prevention program has been established together with the police, rescue services, psychiatrists, social services, and emergency operators to quickly try to prevent an ongoing suicide attempt (Beskow and Nyberg 2016). Further studies on prevention of attempted suicides in traffic seem justified.

Treatment

In total, there were 17 fatalities with elevated levels of toxic emissions in the blood. The most widely used antidote against HCN poisoning is hydroxocobalamin, marketed under the name Cyanokit®. The antidote should be administered intravenously as soon as possible after suspected HCN poisoning (Hall et al. 2009), i.e., preferably at the crash site. In a majority of the present cases, the individual had already died or was too badly injured to be significantly helped by administration of an antidote. However, since the HCN blood levels in some cases were as high as 1.7 µg/g, well above levels for clinical manifestations of poisoning (Chaturvedi et al. 2001), it cannot be ruled out that some cases might have derived a benefit from on-scene antidote administration. This implies that it should at least be considered to have Cyanokit® in ambulances when called to a car incident where a fire is suspected; this would be particularly important inside a confined space such as a road tunnel, where emissions could accumulate—a fast intervention with an antidote could reduce both acute symptoms and long-term effects.

Future studies

A study of emergency services personnel understanding of the toxicity of burning vehicle smoke, as well as preparedness with antidotes, may be of value. Since there is no rapid test that can be used on a crash site to detect blood levels of HCN, it should be included in the autopsy of all fatalities in burning vehicles to better understand the extent of the problem. This study showed that the age of crashed vehicles increased by 0.26 years per calendar year, suggesting that even if new models of vehicles coming out of the factories should become safer regarding fire-related crashes, real world benefits may be delayed. A follow-up study should be conducted in the years to come regarding toxicological patterns of burning electric vehicles (possibly compared to non-electric and hybrid vehicles). Future studies with a control group of people who died in a non-fire car crash are possible to undertake using the In-Depth Studies Database—this was outside the limited scope and aim of the present study.

Limitations

Due to a cross-sectional data structure, we cannot state any conclusions about causality; only about correlations. Unknown raw data confounders may skew the results: importantly, we have not been able to determine if there are confounding regional or local variations (in different parts of Sweden) regarding how the autopsies were performed, e.g., substances quantified.

Acknowledgements

The authors would like to thank Carina Teneberg and colleagues at The Swedish Transport Administration for help with gathering data from the In-Depth Studies Database; associate professor Jonathan Gilthorpe, Umeå University Medical Programme, for kind advice; and the Swedish Civil Contingencies Agency and Umeå University Medical Programme for financial support.

Additional information

Funding

This work was supported by the Swedish Civil Contingencies Agency (DNR 2019-11351), and by the Umeå University Medical Programme.